The smartest machines can learn from nature...
Even though they have relatively small brains, insects are able to learn about their environments remarkably quickly.
Discovering more about the subtle strategies they use enables us to design new kinds of robots with navigation techniques inspired by nature.
At the Centre for Computational Neuroscience and Robotics, we are exploiting the interfaces between biological and computational sciences. By studying the navigation techniques of insects we are advancing biological knowledge as well as developing new kinds of robotic systems. We are also studying information processing in nervous systems. The data collected enable us to build computational models to test ideas about how insects learn. With this knowledge we are able to develop artificial systems for use in robot control and pattern recognition.
By studying how insects operate we are beginning to develop robots for use in planetary exploration, as well as in industry, forestry and agriculture.
Cross-disciplinary work in the Centre for Computational Neuroscience and Robotics
Artificial intelligence (AI) and neuroscience are two areas in which the Â鶹´«Ã½ÉçÇøÈë¿Ú has a long-established reputation for world-class research. In the early 1990s the Evolutionary and Adaptive Systems group became a leading force in biologically inspired AI and artificial life research. To develop the growing collaborations between this group and the Sussex Centre for Neuroscience, the Centre for Computational Neuroscience and Robotics was formed as a joint enterprise.
The aim of the Centre is to exploit the interfaces between the biological and computational sciences. The benefits of a cross-disciplinary approach are reflected in the nature of its research.
Some of the work in the Centre can be characterised as biologically inspired engineering. This includes new approaches to AI, adaptive robotics and electronics. Other research involves the modelling and analysis of biological topics, such as the study of navigational strategies employed by insects in order to both advance biological knowledge and develop new kinds of robotic systems. Insects are able to learn about their environments remarkably quickly, especially considering their relatively small brains. By combining traditional behavioural experiments with computational and robotic modelling we are learning more about the subtle strategies that insects use. This knowledge informs and inspires our design of new kinds of robot control systems which, we hope, will allow us to develop robust robots, able to navigate in uncertain environments.
Other work includes the research on information processing in nervous systems. Specialist equipment enables us to study signals flowing around cultured neuronal networks. The data collected are used to build computational models to test ideas about how information is processed in insect nervous systems, for instance how learning takes place. The insights provided help us develop artificial neural systems for use in a variety of applications including robot control and pattern recognition.
Practical applications for this type of research include robots being used in planetary exploration, and some of our work has been funded by the European Space Agency. Our work may also lead to the production of autonomous industrial robots that are able to find their way around cluttered, changing environments, such as factory floors, or those found in forestry and agricultural applications.
